Function SkipUTF8Chars

util/charset/utf8.cpp:111–124

extern const wchar32 BROKEN_RUNE = 0xFFFD;

static const char* SkipUTF8Chars(const char* begin, const char* end, size_t numChars) {
    const unsigned char* uEnd = reinterpret_cast<const unsigned char*>(end);
    while (begin != end && numChars > 0) {
        const unsigned char* uBegin = reinterpret_cast<const unsigned char*>(begin);
        size_t runeLen;
        if (GetUTF8CharLen(runeLen, uBegin, uEnd) != RECODE_OK) {
            ythrow yexception() << "invalid UTF-8 char";
        }
        begin += runeLen;
        Y_ASSERT(begin <= end);
        --numChars;
    }
    return begin;
}

TStringBuf SubstrUTF8(const TStringBuf str Y_LIFETIME_BOUND, size_t pos, size_t len) {
    const char* start = SkipUTF8Chars(str.begin(), str.end(), pos);